Global Multi-Faceted Application and Evaluation of Three Commonly Used NDVI Products for Terrestrial Ecosystem Monitoring

The Normalized Difference Vegetation Index (NDVI) is a fundamental metric for monitoring terrestrial ecosystem dynamics and assessing ecological responses to climate change. However, uncertainties persist across NDVI products, and a comprehensive assessment of their consistency is lacking. This study conducts a multi-faceted evaluation of three NDVI products, GIMMS V1.2 NDVI (NDVI3g+), PKU GIMMS NDVI (NDVIpku), and MODIS NDVI (NDVImod), to elucidate their performance across ecosystem applications. Our analysis encompasses a comparative analysis of NDVI values, trends, sensitivity to root-zone soil moisture (RSM), and performance in tracking photosynthesis benchmarked against solar-induced chlorophyll fluorescence (SIF). Our results reveal that NDVI3g+ deviates notably from NDVIpku and NDVImod over cold climates and Evergreen Broadleaf Forest (EBF). Additionally, NDVI3g+ exhibits significant global browning, in contrast to the significant greening observed for NDVIpku and NDVImod. Although their responses to RSM are generally uncertain, consistent positive responses appear in Drylands, with NDVImod showing the highest sensitivity. Additionally, the three NDVI products have high seasonality consistency with SIF, except over EBF, and NDVIpku and NDVI3g+ achieve the highest and lowest overall anomaly consistency with SIF, respectively. Furthermore, converting NDVI3g+, NDVIpku, and NDVImod to the corresponding kernel NDVIs improves seasonality consistency with SIF across 85% of the globe.